Index

Show enters and exits. Hide enters and exits.

00:01:18nitay leaves the room.
00:03:16wmoxam leaves the room.
00:03:23heycarsten leaves the room.
00:03:23binary42 leaves the room.
00:06:04rue leaves the room.
00:08:47pauldix leaves the room.
00:13:43jackdempsey leaves the room.
00:24:10explody leaves the room.
00:32:08benburkert leaves the room.
00:32:12benburkert enters the room.
00:32:48Fullmoon_ enters the room.
00:36:57benburkert leaves the room.
00:37:02jackdempsey enters the room.
00:37:06cremes leaves the room.
00:38:01yasuhito leaves the room.
00:40:59benburkert enters the room.
00:44:13dysinger leaves the room.
00:46:45imajes enters the room.
00:47:27binary42 enters the room.
00:51:22Fullmoon leaves the room.
00:53:35stepheneb leaves the room.
00:55:59lopex enters the room.
00:58:54trythil enters the room.
01:00:39trythil leaves the room.
01:02:24jackdempsey leaves the room.
01:07:20yugui enters the room.
01:10:03imajes leaves the room.
01:16:52viimrles leaves the room.
01:24:56imajes enters the room.
01:37:10imajes leaves the room.
01:38:27hemulen leaves the room.
01:52:00jtoy enters the room.
01:53:15sudoer enters the room.
01:54:16dfg59 leaves the room.
02:08:19lopex leaves the room.
02:10:27jtoy leaves the room.
02:14:13blakewatters leaves the room.
02:21:24jackdempsey enters the room.
02:24:40VVSiz_ enters the room.
02:31:00Fullmoon_ leaves the room.
02:33:15blakewatters enters the room.
02:34:00stepheneb enters the room.
02:39:17ezmobius leaves the room.
02:41:53VVSiz leaves the room.
02:42:48trythil enters the room.
02:47:18ijcd leaves the room.
03:01:59stepheneb leaves the room.
03:05:29benburkert leaves the room.
03:11:24c0sin leaves the room.
03:14:36Taggard enters the room.
03:15:49stepheneb enters the room.
03:20:18wmoxam enters the room.
03:23:14benburkert enters the room.
03:24:15antares_ leaves the room.
03:28:37fbuilesv enters the room.
03:33:45rphillips leaves the room.
03:38:59benburkert leaves the room.
03:40:48benburkert enters the room.
03:43:45lchin enters the room.
03:44:08antares leaves the room.
03:46:02yugui leaves the room.
03:51:12seydar enters the room.
03:52:30seydarhowdy, y'all
03:54:03fbuilesvseydar: http://rubinius.lighthouseapp.com/projects/5089/tickets/541-macosx_deployment_target-misnamed-on-m ac-os-10-4
03:54:13fbuilesvcan't reproduce on 10.4 either :(
03:54:27seydaryou dont have the bug?
03:54:29seydaror you do?
03:54:37fbuilesvI don't have it
03:54:41seydarwow
03:54:52seydar10.4.11?
03:54:57seydarPPC?
03:55:02drbraindo you have the latest xcode?
03:55:03fbuilesvIntel
03:55:43seydarfbuilesv: thats our issue. it apparently *only* happens on Tiger/PPC
03:55:43fbuilesvand I'm using the XCode bundled with the 10.4 CD with I think one upgrade from Software Update
03:55:50fbuilesvah damn
03:55:51seydarim using GCC
03:56:24drbrainxcode contains gcc
03:56:32fbuilesv^^
03:56:50seydaryea, but i *think* i installed a new version of gcc. cant remember though
03:56:55seydarwould that be an issue?
03:57:18fbuilesvno idea, is the bug even an issue? I mean, can't you build because of it or something?
03:57:32seydarno, i just set the shell variable
03:57:36seydarand it works like a charm
03:58:01fbuilesvwithout setting it by hand it fails?
03:58:49drbrainthere's somebody here that's more versed in the setting of this env var
03:58:55drbrainDefiler or evan, I think
04:05:39seydarits an issue in config.rake
04:05:46seydarand i'd like to find a way to determine it
04:06:35fbuilesvseydar: the only mentions I found are in libltdl
04:06:49fbuilesvnot sure if that might be setting it
04:08:10boyscout5 commits by Brian Ford
04:08:11boyscout * Updated more CI excludes.; a424ed8
04:08:12boyscout * Converted shared specs to new RSpec-compatible form.; b0a7ffa
04:08:12edwardam leaves the room.
04:08:13boyscout * Updated CI tags for updated specs.; 2fe0593
04:08:14boyscout * Updated RubySpec source to b897f30c.; 08e3566
04:08:15boyscout * Updated MSpec source to 1c3ee1c8.; 946d56e
04:08:32edwardam enters the room.
04:08:32brixencrosses his fingers
04:08:57seydargratz on the commit
04:09:09brixenthe fruits of almost 12 solid horas on those damn specs today
04:11:28seydarbeautiful
04:11:51seydarAND you got paid for it
04:18:33seydar leaves the room.
04:22:25brixenhmm, poor rubuildius_ey64 is all alone now
04:23:17brixendrbrain: are bots running that report to ci.rubini.us?
04:23:22rubuildius_amd64 enters the room.
04:24:11drbrainthe rubuildius ones are not
04:24:31drbrainevan and zenspider know better what's running where
04:24:45drbrainI only performed a few sanity-check duties :(
04:25:15brixenah ok
04:26:27drbrainI think there's only one bot running, maybe?
04:28:13brixenthe ey64 bot appears to be hung on /cgi/html4fr/element_init_spec.rb
04:28:20brixenlooks like amd64 just joined
04:28:36brixenwhatever reports to ci.rubini.us doesn't appear to be reporting
04:33:15drbrainor, the cron hasn't run
04:33:39brixenah yeah, could be
04:36:17zenspider leaves the room.
04:39:32seydar enters the room.
04:53:36seydarRDocing files is as simple as cut & pasting the stuff on ruby-doc.org, right?
05:01:34brixenseydar: should probably use the rdoc in the MRI source files
05:02:04seydarwhy?
05:02:13seydarnot to seem rude...
05:02:36rubuildius_ey64 leaves the room.
05:03:37rubuildius_ey64 enters the room.
05:03:53brixenbecause it's in rdoc format?
05:04:26seydarhm.... excellent point, young sir
05:04:32brixenheh
05:04:58seydarwell.... file.rb will just be oddly doc'd
05:05:09seydarbecause i did the whole thing, and it was good hard work
05:07:45seydarevan: you were right. south park IS awesome
05:14:31rubuildius_ey64Brian Ford: a424ed86e; 2585 files, 8511 examples, 28539 expectations, 0 failures, 0 errors
05:15:11brixenrunning specs really eats memory on 64bit platforms
05:17:31seydarbrixen: is it an MSpec issue?
05:18:09brixennot really
05:18:21brixenit's a too many objects issue
05:19:45seydarso the tests themselves are icky?
05:19:55seydarif the tests are sorta smallish (i assume they are)
05:20:10seydarthen wouldn't the GC kill the excess objects every 8mb?
05:21:11seydarso if one test doesnt go above 8mb, then it shouldn't go above 8mb
05:21:28seydarsomeone tell me where im wrong
05:21:36rphillips enters the room.
05:23:40brixenit seems to me (unsupported by evidence yet) that the GC might no be reclaiming some objects
05:24:14seydaroh wait, my claims were based on MRI GC
05:24:28seydarso there is a bug in the Rubinius GC, i take it?
05:25:27seydari s0 want to get into the GC. its just so... cool
05:27:35ijcd enters the room.
05:27:36jicksta enters the room.
05:27:39seydarbrixen: funny.... gc.c is empty
05:30:11seydarevan: whats up with the gc.c being empty?
05:31:33seydaris it being replaced in object_memory.c?
05:31:55brixenseydar: it's an empty vessel for your soul
05:32:01brixenyou can pour yourself into rubinius
05:32:15seydarthen... i'll fill it with comments
05:32:31seydarand in a month when i return, evan's gonna be like 'wtf'
05:35:15stepheneb leaves the room.
05:38:25edwardam_ enters the room.
05:38:45seydarbrixen: can i actually do that?
05:39:17edwardam leaves the room.
05:40:18brixenseydar: do you like detective work?
05:40:24seydaryea!
05:40:42brixenwell, something you could work on, use lsof to ensure that the specs close all files they open
05:41:07brixenand ensure that all temp files are created with tmp("some_name")
05:41:31brixenfiles including sockets
05:41:53brixensockets closed that is
05:43:22seydaralrighty
05:43:27seydarcan't get to work on it now, though
05:43:47seydarso i'll try to sneak in some work tomorrow. else it'll be in *exactly* 1 month
05:44:01seydarso i'm gonna commit this vessel and head out
05:44:17brixenok
05:44:37brixenwhy *exactly* 1 month?
05:44:53seydarhiking in alaska :-)
05:45:01brixenahh cool!
05:45:06seydaryea! im psyched
05:45:23brixenany mountaineering or just hiking?
05:45:37seydarhiking, dont think any mountaineering. there might be
05:45:50brixencool
05:45:55jackdempsey leaves the room.
05:46:04brixenbeautiful country, from the pictures anyway :)
05:46:30seydari'll take tons of pictures
05:46:40brixensweet
05:47:25boyscout1 commit by Ari Brown
05:47:26boyscout * poured my soul into rubinius, according to brixen; 70a7114
05:47:55brixenheh
05:48:08seydarits part of NOLS, and i'm supposed to come back wildly more mature and have epiphanies. so i need to try not to think about rubinius.
05:48:11seydaralrighty
05:48:14seydargnight brixen.
05:57:18rubuildius_amd64Ari Brown: 70a711483; 2585 files, 8511 examples, 28539 expectations, 0 failures, 0 errors
06:03:01vintrepid enters the room.
06:17:01loincloth enters the room.
06:19:24benburkert leaves the room.
06:19:30rubuildius_ey64 leaves the room.
06:20:23zenspider enters the room.
06:20:24edwardam leaves the room.
06:22:23rubuildius_ey64 enters the room.
06:27:17wmoxam leaves the room.
06:28:51boyscout1 commit by Brian Ford
06:28:52boyscout * Tagged unstable Resolv spec that intermittently hangs on 64bit.; 12312ef
06:33:29nicksieger leaves the room.
06:42:08rubuildius_amd64Brian Ford: 12312ef2a; 2585 files, 8510 examples, 28537 expectations, 0 failures, 0 errors
06:43:08yugui enters the room.
06:43:20rubuildius_ey64Brian Ford: 12312ef2a; 2585 files, 8510 examples, 28537 expectations, 0 failures, 0 errors
06:43:21rubuildius_ey64Ari Brown: 70a711483; 2585 files, 8511 examples, 28539 expectations, 0 failures, 0 errors
06:49:43atmos enters the room.
06:51:15yugui leaves the room.
06:57:40thehcdreamer leaves the room.
07:03:56rue enters the room.
07:12:35roo enters the room.
07:13:15trythil leaves the room.
07:17:59mkrauskopf enters the room.
07:28:30rue leaves the room.
07:37:05krsh enters the room.
07:38:33rueMorning
07:44:02vintrepid leaves the room.
07:53:36benny enters the room.
07:55:21brixenmorning rue
07:55:30joachimm enters the room.
07:55:35rueEvening
07:58:22ijcd_ enters the room.
08:04:14ijcd_ leaves the room.
08:04:14ijcd_ enters the room.
08:04:14rueGetting on the boat, so I will probably be working offline at times until I get back to land
08:08:20ijcd_ leaves the room.
08:08:39ijcd_ enters the room.
08:10:11yugui enters the room.
08:12:09mkrauskopf leaves the room.
08:12:15wvdschel enters the room.
08:13:33ijcd_ leaves the room.
08:14:06ijcd_ enters the room.
08:17:46ijcd_ leaves the room.
08:17:48thehcdreamer enters the room.
08:22:31ijcd leaves the room.
08:31:27gnufiedofftopic warning:
08:31:56gnufieddigit{3} @add_protocol_code space '<'; isn't capturing three digits in ragel, why?
08:32:22gnufiedwhere as digit+ @add_protocol_code space '<'; does capture it
08:34:30joachimm leaves the room.
08:36:49roo enters the room.
08:37:46loincloth leaves the room.
08:39:30joachimm enters the room.
08:41:23blakewatters leaves the room.
08:42:17rue leaves the room.
08:44:34BlackEdder enters the room.
08:46:25BlackEdder enters the room.
08:54:58Yurik leaves the room.
09:10:38joachimm leaves the room.
09:12:16octopod enters the room.
09:21:14roo leaves the room.
09:34:56yugui leaves the room.
10:04:12enebo leaves the room.
10:14:33headiushey there, question of process, not strictly rbx related
10:15:13headiusif someone commits a change that causes a regression, do you generally git revert it? I've got a couple regressions on JRuby, and I'm trying to decide if revert is clean/kosher as part of the process
10:17:04headiushmmm, dead calm
10:21:19wmoxam enters the room.
10:32:07michalw enters the room.
10:39:48headius leaves the room.
10:41:07Maledictus enters the room.
11:10:44gnufied leaves the room.
11:13:45chris2 enters the room.
11:16:18sudoer leaves the room.
11:17:59michalw leaves the room.
11:18:34mkrauskopf enters the room.
11:20:16jero5headius: we have a strong pre-commit review process through lighthouse, so we rarely have regressions
11:20:33michalw enters the room.
11:28:58imajes_office leaves the room.
11:32:04imajes enters the room.
11:42:18Fullmoon enters the room.
11:48:27imajes_ enters the room.
11:48:57krsh leaves the room.
11:52:12imajes_ leaves the room.
12:01:23pauldix enters the room.
12:02:09Maledictus leaves the room.
12:38:38Fullmoon leaves the room.
12:45:20yroc enters the room.
13:05:13tim_w enters the room.
13:16:52foysavas leaves the room.
13:21:30Fullmoon enters the room.
13:36:10stepheneb enters the room.
13:40:18yroc leaves the room.
13:44:47cremes enters the room.
13:46:51brainopia enters the room.
13:47:21foysavas enters the room.
13:48:39stepheneb leaves the room.
13:49:50yasuhito enters the room.
13:53:22benburkert enters the room.
13:57:26benburkert leaves the room.
14:05:33yasuhito leaves the room.
14:11:23jtoy enters the room.
14:11:30jackdempsey enters the room.
14:20:25jackdempsey leaves the room.
14:24:04hemulen enters the room.
14:27:39seydar enters the room.
14:28:09brainopia leaves the room.
14:31:03wmoxam leaves the room.
14:36:38binary42 leaves the room.
14:37:12Yurik enters the room.
14:40:33gnufied enters the room.
14:59:28moofbong enters the room.
15:06:22yugui enters the room.
15:08:39binary42 enters the room.
15:14:44jackdempsey enters the room.
15:16:30jtoy leaves the room.
15:16:37wmoxam enters the room.
15:20:03obvio171 enters the room.
15:20:10obvio171 leaves the room.
15:22:02jackdempsey leaves the room.
15:29:01michalw leaves the room.
15:29:11michalw enters the room.
15:30:08krsh enters the room.
15:30:10anonuser leaves the room.
15:33:11jtoy enters the room.
15:37:57heycarsten enters the room.
15:50:47madsimian enters the room.
15:51:54madsimian leaves the room.
15:55:51lchin leaves the room.
16:00:14wvdschel leaves the room.
16:16:12jtoy leaves the room.
16:18:17lopex enters the room.
16:26:31Fullmoon leaves the room.
16:38:40gnufied leaves the room.
16:38:59seydar enters the room.
16:39:40jackdempsey enters the room.
16:41:43brixenhmm, the calm before the storm?
16:42:33Defilerlistens for the breeze
16:42:38brixenheh
16:42:49Defilerbecalmed! yarr!
16:43:29brixenhah, I shall stir the cauldron and summon the daemons
16:46:12seydarbrixen, you are just full of deep, often fantasy-related, pearls of wisdom
16:46:22mkrauskopf leaves the room.
16:46:33TheProkrammerIf MacBeth shows up, we're screwed.
16:46:45Defilerluckily the woods haven't moved
16:47:08seydarplus, we consider C-sections to be actual births now
16:47:23seydaroh wait, that would make macbeth invincible
16:47:29TheProkrammerright.
16:47:30DefilerMuhahaha
16:47:44DefilerIt would have to be a chick fighting him then
16:47:45TheProkrammerThough, that "Man" that had the baby girl, may nullify that.
16:48:56brixenfresh fun with compiler bugs: http://rubinius.lighthouseapp.com/projects/5089-rubinius/tickets/662
16:49:55dfg59 enters the room.
16:50:31seydarbrixen: MRI doesn't evaluate the body either
16:50:39seydaroh wait, exception. right.
16:50:47brixenseydar: you might enjoy these pearls of wisdom more than mine own: http://www.brainyquote.com/quotes/authors/e/edsger_dijkstra.html
16:51:06seydarbrixen: dijkstra's a cool dude. but wow, these are awesome
16:51:15brixenseydar: but you might want to shield your eyes when looking at the words of a god among mere mortals
16:51:46seydaryou forget the fact that he's dutch
16:51:52brixenoh?
16:52:14brixena dutch god among heathens?
16:52:26brixenconfuses his cultures and religions endlessly
16:52:46seydarcorrect
16:53:01seydaryou excited for D&D 4e?
16:53:26brixenheh, no idea what that is
16:53:43brixenlove this: "Object-oriented programming is an exceptionally bad idea which could only have originated in California."
16:53:44nexcastellanI already reported but 662 as 660. :)
16:53:52brixennexcastellan: oh good!
16:54:06brixenflagellates himself
16:54:29seydarbrixen: dungeons and dragons? have you never been a geek?
16:54:34brixennexcastellan: I shall review the stellar work of comrades in less sleep-induced haze in the future
16:54:40nexcastellan:) Hey, normally I just bring it up here, I'm only just starting to get good at also reporting on lighthouse.
16:54:47brixennexcastellan: yep, thanks
16:54:58brixenseydar: I only pretend to be a geek online
16:55:11seydaryou should fix that
16:55:26brixennexcastellan: I do like my formatting better though :D
16:55:41nexcastellanHahaha.
16:56:02brixennexcastellan: but I fixed up yours, so it's all good now :)
16:56:54seydarcan i guess the issue at heart here?
16:57:28brixenseydar: speak to me in code
16:57:35seydar# the () should make it obvious that there is an expression there
16:57:59seydar# so i think it's that an empty set of parenthesis are not considered an expression
16:58:01hemulen leaves the room.
16:58:10brixenheh
16:58:39Defilerthat is funny as hell
16:58:43nexcastellanBy the way, I have an "amusing" issue I'm trying to track down in Rubinius. It seems that Rubinius sometimes gets confused with a Thread class inside a module unrelated to the main Thread class. I'm seeing that sometimes a Thread.current[] call is trying to reference the wrong class. I can't yet isolate the problem to a nice test case.
16:58:52BlackEdder enters the room.
16:59:04DefilerYeah, I guarantee we have bugs like that currently
16:59:16DefilerAny reproducing code would be greatly appreciated
16:59:26Defilerhot, sweaty pics of code
16:59:40seydarwhoa there Defiler. keep it pg
16:59:42nexcastellanI wasn't able to get reproducing code yesterday though I certainly tried. Will try today as well.
17:00:04nexcastellanOh great, Defiler you just made me spill coffee on myself. :)
17:01:38seydarwhere in the compiler does it describe what an expression *is*? I'm looking at nodes.rb, but its not very helpful
17:04:12Defilerthe compiler only implements nodes for things the parser hands it
17:04:19Defilerso that is where the definition of 'expression' lies
17:04:32Defilerwhere none bethink to tread
17:05:22brixenseydar: you can compare e.g. echo "if (2) then 1 end" | bin/rbx describe with echo "if () then 1 end" | bin/rbx describe
17:05:26seydari shall do it, sir!
17:05:27brixenand look at the sexp
17:05:44seydaralthough, i do not know the way...
17:07:03jackdempsey leaves the room.
17:07:36TheProkrammerYou have my keyboard!
17:07:39seydarwhat does describe do again?
17:08:29brixenseydar: it lists the sexp and the bytecode
17:08:42seydargotcha
17:08:51seydaryea, so the bug lies within the sexp parser
17:08:52brixenseydar: did you type the two commands in your shell?
17:08:57seydarbrixen: yea
17:09:00brixenok
17:09:58seydarrun this: echo "if \n then 1 end" | bin/rbx describe
17:10:59cremesnexcastellan: fyi, i believe the dramatis gem exposes some threading bugs in rubinius where the thread @owner is missing; i saw that a few days ago when playing around
17:11:19cremesit's probably possible to build a reproducible case from that code
17:12:34nexcastellanHhhhmm. thread.rb seemed to have some rather dubious code that would certainly lead to problems.
17:12:46nexcastellanWhat is the dramatis gem for?
17:13:36nexcastellanThe Queue and SizedQueue are more likely to work now, but mostly on the basis that they had almost 0% chance of working before.
17:13:45seydarnexcastellan: actor implementation
17:14:26seydarbrixen: did you run the snippet?
17:15:07seydarfalse alarm
17:15:19stepheneb enters the room.
17:15:19seydarit was reading the \n as though it was actually \ and a n
17:15:24Maledictus enters the room.
17:23:32seydarhmm... so we're looking for syd_add_to_parse_tree in the code
17:23:54seydarin grammar.y, its defined as a function thats supposed to exist yet has no body.
17:25:38eventualbuddha enters the room.
17:25:46octopod leaves the room.
17:26:05headius enters the room.
17:27:15seydarthis is killing my brain
17:29:21headiusdeath to brain
17:30:14seydarwell said. grammar_runtime.c is the culprit
17:30:26headiuswazzat
17:30:34headiusparser-related I presume
17:30:41seydarbut i have no skill points in decipher (heiroglyphics)
17:31:02seydarheadius: yea. run: if () then 2 end
17:31:16seydarhttp://rubinius.lighthouseapp.com/projects/5089-rubinius/tickets/662
17:32:03brixenseydar: sorry have to run to class, but I leave you in capable hands
17:32:18seydarhah! nice one
17:32:19headius() should just be a nil
17:32:25seydaryea
17:32:29brixenheadius: btw, your earlier Q, we typically revert unless the fix is easy
17:32:32seydarbut it doesn't even parse
17:32:44seydarbrixen: ask two good questions for me.
17:33:07headiusbrixen: cool, thanks...I figured as much since it's just as easy to revert a revert and the twiddle it for modifications
17:33:25brixenseydar: not sure what ya mean. but I did compare the output to parse_tree_show and it seems rbx sexp is missing the [nil]
17:33:37headiusseydar: hmm, looks like it parses here
17:33:51headiuserror I'm seeing comes in the compiler
17:33:52brixenseydar: or perhaps our sexp is just different for that construct
17:34:35seydarso the error is in the compiler, we agree on that?
17:34:40brixenyeah, compiler is not getting what it expects, that's the visible error
17:34:57brixenbut I'm not saying what the root cause is
17:35:02brixencus I don't know ;)
17:35:18brixenbbl...
17:35:36seydarlater alligator
17:35:52seydarheadius: could the problem be that () isn't getting parsed as an expression?
17:36:02seydaror its getting parsed as an improper expression?
17:36:27seydarits obviously getting parsed as something, because it results in the 'then' part as getting its good
17:36:30seydargoods*
17:36:41headiusseydar: ultimately I doubt it's the parser since you guys just use ruby's parser
17:36:52headiusbut PT is probably outputting wrongness
17:37:00headiuscheck PT under regular ruby and see what you get from it
17:37:01seydarPT?
17:37:04headiusparsertree
17:37:05headius-r
17:37:35NoKarma enters the room.
17:38:05seydarkk
17:38:15seydari *thought* that's what i was looking at
17:38:23seydarbut clearly im not
17:38:39seydarfollowing File::to_sexp, i got here
17:39:19massirony: httpd.apache.org is down
17:39:40Defilerat last!
17:39:40hemulen enters the room.
17:40:11seydarheadius: syd_add_to_parse_tree has comments that talk about sexp... so i think im in the right spot
17:40:13Defilerheadius: We generally use the 'real deal' "git revert" command, yes
17:40:15edwardam enters the room.
17:40:45headiusDefiler: danke
17:41:26Defilerwhee a time capsule arrived
17:41:28Defilershould be fun
17:42:21seydarok. line 248 begins the mess of hell (i think)
17:42:22masshint: sync over a physical cable
17:42:36massat least the first time
17:42:46seydarthrow in some printf's and we should be able to see where this is screwing up
17:42:53seydarunfortunately, i have to leave
17:43:16seydarsomebody want to take it over?
17:43:47blakewatters enters the room.
17:44:53thehcdreamer leaves the room.
17:45:08eventualbuddha leaves the room.
17:45:15seydarawesome, thanks for stepping up Defiler.
17:45:21seydar leaves the room.
17:50:47tim_w enters the room.
17:50:52BlackEdder leaves the room.
17:54:06tim_w leaves the room.
17:59:30mkrauskopf enters the room.
18:04:57headiusyou guys might find this interesting: http://mail.openjdk.java.net/pipermail/porters-dev/2008-July/000160.html
18:05:02fbuilesv leaves the room.
18:05:04headiusin a roundabout way
18:16:15gnufied enters the room.
18:17:39dysinger enters the room.
18:17:52massheadius, what is going on with java 7 anyway? It seems to have no formal featureset or schedule
18:18:31tim_w enters the room.
18:18:55antares enters the room.
18:20:47krsh leaves the room.
18:24:15ijcd enters the room.
18:25:46headiusmass: oh it's moving along
18:26:01headiusI'm almost exclusively interested in the dynamic invocation work, naturally
18:26:18headiusjava-speed dispatch for ruby
18:27:29eventualbuddha enters the room.
18:28:33eventualbuddha leaves the room.
18:32:54benburkert enters the room.
18:34:48massheadius: the process is open but not well summarized, I got the feeling that jdk7 was a misnomer because a lot of work was going into repackaging and improving 6
18:35:10massso its more like jdk8
18:36:17headiusI think largely it's less concrete because of openjdk...there's a lot of challenges adapting this process and that large a code base to a new fully OSS model
18:36:26headiusalso, rather than having Sun occasionally toss updates and announcements over the wall, most discussion now happens on public lists
18:36:39headiusso I think it's just different
18:39:23massI'm still sorta expecting a JSR on features to come out at least half a year before a sun final binary hits their site
18:39:31massI just want to start being able to require java 6 :)
18:39:34moofbong_ enters the room.
18:50:11pauldix leaves the room.
18:53:00joachimm enters the room.
18:55:23moofbong leaves the room.
18:57:34c0sin enters the room.
19:00:56joachimm leaves the room.
19:02:40ezmobius enters the room.
19:04:56benburkert leaves the room.
19:07:59michalw leaves the room.
19:09:13benburkert enters the room.
19:19:00binary42 leaves the room.
19:19:43headiusmass: heheh...it's probably safe to do that right now, don'tcha think?
19:20:11headiusbut we still support java 5 too largely for that reason...always need to support at least one major java version back
19:21:37stepheneb leaves the room.
19:22:09stepheneb enters the room.
19:22:47stepheneb leaves the room.
19:23:06stepheneb enters the room.
19:26:15fbuilesv enters the room.
19:28:48joachimm enters the room.
19:29:55cremesheadius: if you drop java5, that orphans anyone using OSX on PPC; there is no java6 release for ppc
19:32:52headiusno plans to drop java 5 any time soon
19:32:57dfg59cremes: is there a java6 release for intel macs?
19:33:11headiusjava 5 and java 6 are nearly identical as far as we're concerned (API-wise)
19:33:33cremesdfg59: yessir, apple released it about 6 weeks ago
19:33:48cremesheadius: just an fyi... :)
19:33:54headiusonly for 64-bit intel macs though
19:33:59headiussucks 2 b me
19:34:19dfg59cremes: how can i install it? i don't get it automagically through software update do i?
19:34:46headiusif you want java 6 on mac but have 32-bit, you can use soylatte
19:35:52cremesdfg59: i don't recall; i *think* it came through via software update; what kind of mac are you on?
19:36:05blakewatters leaves the room.
19:36:37dfg59cremes: hrmm, I'm on a mbp
19:36:57cremesand what release of osx?
19:37:03cremesyou need leopard
19:37:04dfg59cremes: java -version tells me 1.5.0_13
19:37:18dfg59i'm on 10.5.4
19:37:22cremesokay, let's take this chat private
19:37:44headiusyeah, not particularly relevant to #rubinius, at least not until we port it :)
19:38:40pauldix enters the room.
19:42:23thehcdreamer enters the room.
19:46:19benburkert leaves the room.
19:47:45chris2 leaves the room.
19:49:51ezmobius leaves the room.
19:52:55pjstadig enters the room.
19:55:15twbray enters the room.
19:57:52rue_XIVHola
19:57:55moofbong_ leaves the room.
19:58:48Defilerroo-e the 14th, eh?
19:58:50Defilerthat's pretty bad
19:59:33rueEh, roo and rhubarb were apparently too difficult to associate with me :P
20:01:15binary42 enters the room.
20:01:15DefilerWhich SHA1 library/function/universe should the new VM use?
20:02:21drbrainI'd use whichever C implementation is in lib/ext/digest/sha1
20:03:24nathansobo enters the room.
20:06:40Defilerwhat about shotgun/lib/sha1.c?
20:08:13tarcierihmmm, don't suppose evan has been around today...
20:08:18tarcierihad a crazy idea to discuss
20:08:23DefilerHe is traveling
20:08:27smparkes1 enters the room.
20:08:27tarcieriaah
20:08:29tarcieriok
20:09:03tarcieriI'm wondering if it'd be possible to do something sort of like the Erlang code server for managing the environments of multiple VMs (and eventually nodes)
20:09:05drbrainwe aren't good enough for your crazy?
20:09:23smparkes1 leaves the room.
20:09:41tarcierii.e. introspect the current environment and message pass the bytecode for all loaded modules / classes to another VM where a server there would load them back in
20:09:53DefilerYou think you are bad ass enough for us not to be able to implement your ideas? ha ha ha the hubris
20:10:03tarcieriheh
20:10:21joachimm leaves the room.
20:11:45drbrainrue probably has half an idea of the possibility of that
20:12:15DefilerIt doesn't sound like a challenge, unless the VM needs to keep running while it is dumping. heh
20:12:25Defilerthat sounds fickle
20:13:39tarcieriit's really just having something that you can talk to using Actor messages which sticks an interface on what modules / classes or loaded and lets you either get or set the associated bytecode
20:17:48moofbong enters the room.
20:19:05twbray leaves the room.
20:27:21ezmobius enters the room.
20:28:44atmos leaves the room.
20:28:47atmos enters the room.
20:31:27nathansobohey do you guys know of any good literature for implementing a VM for an actor-based language? i'm looking for the closest thing i can get to the Blue Book for actor based languages
20:31:42nathansoboi know you guys are doing actors in rubinius
20:31:53nathansobobut you're also doing some OO stuff
20:31:55DefilerMan, I can barely think of one book that could have that description
20:31:57tarcierinathansobo: Actor-based like Scheme, or actor-based like Erlang?
20:31:59nathansobomaybe a stack based VM is ideal for moth
20:32:01DefilerThat is pretty cutting-edge business
20:32:04nathansobolike erlang
20:32:18nathansoboi'm toying with building a scripting language
20:32:21tarcierinathansobo: I don't know of many books on the matter, but there's tons of papers on BEAM / HiPE
20:32:46nathansobothat exposes erlangish semantics but doesn't make my eyes bleed
20:32:48DefilerThere's some event-based stuff in one book I read
20:32:48tarcierinathansobo: If you want to build a scripting language you may as well build it on BEAM / HiPE directly rather than writing your own VM... that's what I'm doing
20:32:50nathansobowould you consider BEAM the state of the art for actor based VMs?
20:32:54DefilerWhich was interesting, but not really the same
20:33:02dlee enters the room.
20:33:11tarcierinathansobo: It's the state of the art for concurrency and distribution
20:33:12DefilerIs there even a second actor-based VM to compare it to?
20:33:12nathansoboi kinda wanna make my own, just for interest's sake
20:33:46tarcieriDefiler: There's nothing else like the Erlang VM to my knowledge
20:33:46nathansoboso BEAM is basically the smalltalk VM of the actor world, to make a rough analogy
20:33:59nathansobocool. any specific papers you'd suggest?
20:34:02tarcieriyeah, BEAM is the canonical Erlang VM
20:34:19nathansoboand perhpas the canonical actor vm, from what you're telling me
20:34:32Defilererlang is the canonical actor language
20:34:43Defilerso I guess its canonical vm would also be the canonical actor vm
20:34:43tarcierilots of people have lots of different ideas about the Actor model
20:34:53tarcierihttp://www.erlang.se/publications/publications.shtml
20:34:58tarcieri^^^ there's plenty of papers available there
20:35:09nathansobocool. thanks a lot guys!
20:35:20nathansoboi look forward to sharing my thoughts with yall as things develop
20:35:29nathansoboafter a taste of erlang i think there's so much more to do with these ideas
20:35:41nathansoboseems to me that the actor model really subsumes OO
20:35:47tarcierinathansobo: you should check out Reia: http://reia-lang.org/
20:35:53nathansoboi did a little
20:35:56tarcierinathansobo: I don't think the two are incompatible
20:36:05nathansoboi just mean it's more general
20:36:28nathansoboi did check out reia a little: is that yours?
20:36:31tarcieriI think Erlang-style concurrency works more like how Alan Kay originally envisioned OO than anything we have presently
20:36:33tarcieriyes
20:36:51nathansobosweet. we should def talk
20:36:53Defilerlots of things about programming still bug me, despite rubinius being the best thing
20:36:54tarciericool
20:37:09tarcieriRubinius does an awesome job with concurrency
20:37:28tarcieriespecially compared to the "stab my eyes out with a fork" implementation in MRI
20:37:39tarcierirb_thread_schedule() is just silly
20:37:40DefilerHas anyone tried Rubinius out with real / semi-real concurrent code?
20:37:50nathansoboyeah i'm wondering if their actor design is a good example for a language based 100% on actors
20:37:54nathansobothe rubinius one
20:38:20DefilerIt would be nice to kick the bugs out as our version number rises
20:38:28tarcieriDefiler: not yet... I'll be converting my Revactor projects over as soon as I can get an "inets" kinda thing going
20:39:13tarcieriDefiler: that is... something which processes incoming network I/O events and converts them to Actor messages
20:39:22Defilercool
20:39:31DefilerThat sounds like a good smoke-test
20:39:43tarcieriyeah
20:39:48Defileran octopus of concurrent networking
20:39:52tarcieriI have a peer-to-peer media streaming server written using Ruby actors
20:39:58tarcieriso that should be a good acid test :)
20:40:03Defilerrise, kraken!
20:40:17headiustarcieri: are you familiar with the "proactor" model for IO?
20:40:24tarcieriheadius: yeah
20:40:31Defileryou're just making up words now
20:40:31tarcieriheadius: I wanted to do some proactors in C for Rev
20:40:42tarcieriheadius: Scheduler can do proactor sort of stuff
20:40:51headiusok...NIO2 is supposed to provide a full proactor setup
20:41:09tarcieriheadius: I think it'd be really nice to add libeio to Rubinius... and have high performance portable async file I/O
20:41:17smparkes1 enters the room.
20:41:26headiusI haven't had a chance to look at the API, but it sounds like NIO2 is going to be the bee's knees
20:41:32tarcierinice
20:41:44tarcieriyeah NIO seemed to get... halfway there :/
20:41:50headiusit was truly halfway
20:42:04headiustoo many of these potentially great APIs get kneecapped due to lack of resources
20:44:44smparkes1 leaves the room.
20:45:42tarcieriI think all you really need are these little observer objects for IO readiness (reactor) or actual IO (proactor)
20:46:04tarcieriand everything follows from there
20:46:06tarcierithat's what Rev does
20:47:04tarcieriI see people on the EventMachine mailing list like "I want to do thing X in the event loop"... some guy's trying to do raw sockets, which I guess Ruby doesn't support... but if you could wrap that into an IO object, Rev could wait for events
20:49:37headiusrev ought to be trivial to build atop NIO then
20:49:43headiusso many projects
20:50:01tarcieriindeed
20:50:02tarcieriheh
20:50:09nathansobo leaves the room.
20:55:20blakewatters enters the room.
20:59:06ruetarcieri: A code server would be useful and should be doable
20:59:20nathansobo enters the room.
20:59:24tarcierirue: excellent
20:59:49tarcierirue: in the end all I *really* want is a mechanism for cloning the current environment on another VM/node
21:00:31antares_ enters the room.
21:00:34seydar enters the room.
21:01:03tarcierirue: basically what I'm thinking is when you create a new VM (or eventually a remote node) it starts the code server and an RPC service
21:01:33benburkert enters the room.
21:01:35rueYeh
21:01:36tarcierirue: then you're free to load whatever you want into the remote VM, and once you've got the environment set up you can use the RPC service to kick off whatever
21:01:53tarcierirue: I'm gonna take another stab at a multi-VM mongrel this weekend
21:01:57rueYep, very handy to have in many cases
21:02:12nathansobowhere can i read about nio2
21:02:22nathansobogoogle doesn't bring up much
21:03:02nathansoboit's a java api?
21:03:10tarcieriyes
21:03:18rueCool, I am doing a little backtracking before getting into the C++ where we can do all the magic distributed stuff
21:04:55tarcierirue: the hard part about a multi-VM Mongrel is getting the environment to the point where you can send messages directing the remote VMs to instantiate the proper HttpHandlers and add them to a VM-local instance of the HttpServer
21:05:24tarcieriif you just had one file to require to pull in the whole environment it'd be fine... that's what I'll do for now I guess :/
21:05:34dbussinkare the cpp tests running slow for other people too?
21:05:47dbussinkor is it the whole llvm overhead that's causing this?
21:06:07benburkert_ enters the room.
21:07:52seydarevan said they were running 2x as fast
21:09:43dbussinkthat was vm performance
21:09:56dbussinkmost of the tests don't really do anything with the vm
21:10:25dbussinkbut the tests are definitely slower then before
21:10:50headiusmaybe there's a startup hit running atop llvm
21:11:04dbussinkyeah, that's what i'm thinking too
21:11:08dbussinka first run performance hit
21:11:21headiusseems like there would be....now tests are going to be parse + compile to rbc + translate to llvm bytecode + jit
21:11:35headiusmore stages in that pipeline than even in jruby
21:12:02dbussinkyeah, but these are just the vm tests themselves
21:12:16dbussinkwhich test the c++ code from the new vm
21:12:24headiusahh, well you're probably still right, llvm stuff has to add something to the end-to-end process
21:12:56smparkes leaves the room.
21:15:41wycats leaves the room.
21:21:40nathansobo_ enters the room.
21:21:44benburkert leaves the room.
21:23:57seydar leaves the room.
21:27:00vjmp enters the room.
21:30:22Maledictus leaves the room.
21:32:18antares_ leaves the room.
21:33:50smparkes enters the room.
21:35:10smparkes leaves the room.
21:35:48seydar enters the room.
21:35:59seydarasdf
21:36:01seydar(whoops)
21:36:42vjmp leaves the room.
21:37:27vjmp enters the room.
21:37:53seydarevan: for the compiler bug, why wouldn't it be when the sexp is formed?
21:37:58nathansobo leaves the room.
21:38:51headiuswhat does parsetree output for that code?
21:39:09smparkes enters the room.
21:39:18seydaris there a pastie bot somewhere?
21:39:28vjmp leaves the room.
21:39:45seydarheadius: http://pastie.org/235969
21:39:45smparkes leaves the room.
21:40:27headiuswhat's the order of args to if there?
21:40:44headiusseems like the condition is missing
21:40:59seydarrefresh it, i update the paste
21:41:16headiusyep, condition is missing
21:41:17benburkert_ leaves the room.
21:41:18headiusseems to be a PT bug
21:41:32seydarthe ruby parser catches it, though
21:41:36jicksta leaves the room.
21:41:42seydarso grammar.y et al is fine
21:41:45headiusoh sure
21:41:53headiusobviously ruby can run this :)
21:41:58headiushere it is in JRuby
21:42:08seydari said maybe grammar_runtime.c, since thats the sexp conversion
21:42:56enebo enters the room.
21:43:01headiusI suppose it could be the compiler
21:43:28seydarwhat do you mean by compiler?
21:43:46headiuswell, I dunno how rubinius compiler works
21:43:53headiusit *seems* to me that that sexp is wrong
21:44:00smparkes enters the room.
21:44:01headiusand that seems to match the error from the compiler
21:44:13headiusbut perhaps the compiler needs to be modified to handle if with only two children
21:44:17smparkes leaves the room.
21:44:25headiusthat would seem odd to me
21:44:50nathansoboare there any good discussion threads about the current state of rubinius's approach to concurrency?
21:44:58headiusI'd expect it to be [:if, nil, [:newline, 1, "(eval)", [:lit, 1], nil]
21:45:14headiusor [:lit, nil]
21:45:21headiusI don't know how pt represents a literal nil
21:45:46seydari think adding support for if to have 2 args is a bad idea
21:46:17seydarbecause i feel like this bug is lower level than that, and PT isn't assigning it to some correct thing
21:46:45vjmp enters the room.
21:47:09seydarbut sure, squash 'em while we can
21:47:45seydarheadius: do you know how to adjust IfNode appropriately?
21:47:48brixenif I read this right, our parse is different than parse_tree 2.1.1 (ours is missing [nil]) http://pastie.org/235972
21:47:58brixenbut I don't know how different our sexp is supposed to be from PT's
21:48:00headiusseydar: not in the slightest :)
21:48:17headiushonestly though, it doesn't seem like a two-arg if sexp is correct
21:48:25seydaraHAH!
21:48:26seydari knew it
21:48:34seydarbrixen: thank you thank you thank you!
21:48:38brixennathansobo_: not really any threads on that per se, but you might google the IRC logs
21:48:50brixennathansobo_: http://www.donttreadonme.co.uk/rubinius-irc/
21:49:21brixennathansobo_: Channels at the bottom, Tasks on top of that, Threads on top of that
21:49:30brixenbut tarcieri can correct me if I'm wrong ;)
21:49:36nathansobofrom what i gather: there's a pi calculus based model for communication between VMs running in different OS threads, and within each of these, there are green threads
21:49:54nathansoboso channels is handling communication between the individual machines
21:50:11benburkert enters the room.
21:50:30brixenoh, there's an primitive IPC between VMs in the multi-VM structure
21:50:50tarcierithere's more than primitive IPC
21:50:55nathansoboare those channels?
21:50:55seydarheadius: how's Duby coming?
21:51:08headiusit's not...maybe after jruby 1.1.3 is out I'll take some time on it
21:51:11tarcieriActors can communicate over the VM message queue
21:51:12brixentarcieri: primitive in the sense of only marshaling primitive types I thought
21:51:16tarcierioh
21:51:22smparkes enters the room.
21:51:56hemulen leaves the room.
21:53:00brixennathansobo_: we've got some specs for VMActor::Container#spawn_actor and a few specs for Actor, but not much :/
21:53:27vjmphi. question about vm test: should they all pass? I got: Failed 4 of 450 tests Success rate: 99%
21:53:28brixennathansobo_: look in spec/library/actor|vmactor
21:53:35nathansobookay
21:53:42brixenvjmp: afaik, there's 4-5 failures atm
21:53:53vjmpok
21:54:01stepheneb leaves the room.
21:54:39smparkes leaves the room.
21:54:51smparkes enters the room.
21:54:56brixennathansobo_: also, lib/actor.rb and lib/actor/filter.rb for your reading pleasure written in Ruby :)
21:55:21vjmpbrixen: thank you
21:55:32nathansoboso tasks are basically a private stack and register set
21:55:34brixenvjmp: n/p
21:56:23smparkes leaves the room.
21:56:50seydarnathansobo_: yea, i think that's right
21:57:19seydarzenspider: ping
21:57:31seydarbrixen: you know how to get PT 2.2 into rubinius?
21:57:45vjmpI see lots of c++ code in vm branch; is it still going to be ruby-in-ruby, or have plans changed?
21:58:14brixenseydar: no, I just parse_tree_show on an .rb file
21:58:41seydarinteresting....
21:58:58brixenvjmp: it's a mixture. the core VM is being rewritten in C++, the core libs are mostly ruby
21:59:35hemulen enters the room.
21:59:42brixenvjmp: we'll see how it goes writing the VM in some sort of Ruby that can be processed to do the low level bit twiddling that is needed
21:59:57seydardibs
22:00:00brixenvjmp: for now, it's a C++ VM with comiler tool chain and core libs in Ruby
22:00:07brixens/comiler/compiler/
22:00:41brixennathansobo_: crap, our rdoc task does not appear to generate for the .c files
22:01:02brixennathansobo_: I think there's some additional code docs on Task, I'm looking
22:01:17nathansoboyou are so gd helpful
22:01:23nathansoboi gg back to pairing with someone
22:01:28nathansobothanks
22:01:31vjmpbrixen: ok. so what about those ruby libs in main branch, is there any point of fixing thing there while c++ vm rewrite is going on?
22:01:34nathansoboi'll keep this open
22:01:39brixennathansobo_: heh, n/p
22:01:53brixennathansobo_: we'll expect the return favor on the Ruby PEG ;)
22:02:13brixenvjmp: absolutely, those ruby libs will go right over
22:02:30brixenvjmp: as soon as the new vm is loading everything
22:03:01nathansoboi feel so daunted by it right now. toying with making up a language at the moment
22:03:06nathansobohence all my questions about concurrency
22:03:14brixencool
22:03:18nathansoboi've come up with a cool sketch
22:03:25seydarnathansobo_: what are you designing?
22:03:25vjmpbrixen: so if there is a failing spec in main branch, it fails in vm too?
22:03:36nathansoboi only have a syntax for it right now
22:03:41nathansobono actual compiler or anything
22:03:46nathansoboi'll pastie something
22:03:49brixenvjmp: not necessarily, the c++ vm isn't to the point of running arbitrary ruby code yet
22:04:22nathansobohttp://pastie.org/235986
22:04:24brixenvjmp: if you're interested in fixing errors in core libs, go for it
22:04:27nathansobocalling it cast
22:04:37nathansoboit's basically a scripting language version of erlang/otp
22:04:45brixenheh eigenmodule
22:04:47nathansobofoo:bar is the cast operator
22:04:52nathansobofoo.bar is a call
22:04:59vjmpbrixen: any knowledge when vm is about to handle ruby code?
22:05:00nathansobomeaning a cast and immediate receive
22:05:04nathansobolike gen_server:call
22:05:21brixenvjmp: WIP, nothing definite but a bunch of concurrent tasks being worked on
22:07:45benburkert leaves the room.
22:07:53vjmpbrixen: ok. btw, what is Ruby PEG? URL?
22:10:17brixenvjmp: well, there's this: http://rubyforge.org/projects/ruby-tp-dw-gram
22:10:38brixenvjmp: but nathansobo_ wrote this http://treetop.rubyforge.org/ and kick around the idea of implementing a Ruby grammar in a Ruby PEG
22:10:52brixenand we kick*
22:12:36brixennathansobo_: when you have time, in the rbx repo, git show be813e67 and 172952d2
22:12:46ljulliar enters the room.
22:14:46seydarnathansobo_: nice job on treetop!
22:15:10ljulliarDoes anybody know how to mimic an MRI hooked variable define at C level?
22:15:41ljulliarFor instance $. (current line in file) is hooked to a setter function
22:15:56ljulliarrb_define_hooked_variable("$.", &lineno, 0, lineno_setter);
22:16:15benburkert enters the room.
22:17:32drbrainljulliar: for $., look in kernel/core/io.rb
22:18:00drbrainin MRI, where you would call the hook, we just set the global
22:18:49ljulliarNo that's not the case
22:19:07ljulliaryou also have to set the gets_lineno variable that is needed in ARGF
22:20:27drbrainwouldn't you ask $. for the current line number and advance to it in ARGF?
22:20:34drbrains/wouldn't/could you/
22:21:43ljulliarLet me see...
22:25:56nathansobothe problem with the notion of writing a ruby PEG is that PEGs can't actually express ruby
22:26:09nathansoboso it's some novel research to figure out how to extend them to do so
22:26:17nathansoboand i got a bit burned out on it for a while
22:26:22nathansobobut i'm feeling the wind again
22:27:12ljulliardrbrain: The problem is when you say $. = 2000 you must set both @lineno from io.rb and @gets_lineno from argf.rbo. They have to be two different variable s
22:27:29ljulliarso you really need to hook a method as setter of $.
22:27:46drbrainok
22:27:56drbrainlet me look, I think there's something for this
22:28:29drbrainkernel/core/global.rb has GlobalVariable#set_hook
22:28:35ljulliarI though of extending the global variable with a method but this doesn't to work very well
22:28:38pauldix leaves the room.
22:29:20drbrainand kernel/core/kernel.rb is where various hooks are defined
22:29:29drbrainwill that help?
22:29:37moofbong leaves the room.
22:30:09mkescher enters the room.
22:30:34ljulliarlet me take a look
22:32:38ljulliarterrific! This is exactly what I need. Thanks!
22:33:01drbrainsure
22:39:07seydari'll see you all in a month
22:39:12seydar leaves the room.
22:39:28smparkes enters the room.
22:40:10drbrainwhere is he going?
22:40:38smparkes1 enters the room.
22:40:53ljulliardrbrain: well... almost there! I couldn't find any exemple of a setter. Do you think one can pass a parameter ?
22:41:09fbuilesv leaves the room.
22:41:15drbrainI would think so
22:41:43ljulliarIt doesn' to try, hey?
22:41:53ljulliarIt doesn't hurt
22:41:55drbrain:)
22:43:25headiusheh
22:43:28headius➔ jruby -rjruby/vm -e "JRuby::VM.spawn('-e', 'puts 1').join"
22:43:28headius1
22:43:47headiusbtw, are the stdin etc on a VM intended to be pipes to the child VM's side?
22:45:02tarcieriyes
22:45:06tarcierior at least, that's how it's implemented now
22:45:25tarcieriit's unfortunate that you can't multiplex them though :/
22:46:03headiuscan't select etc?
22:47:01tarcierioh, maybe it's just mixing read and write that's the issue
22:50:12mkrauskopf leaves the room.
22:50:24headiushmm, that's odd, why would that matter?
22:52:10fbuilesv enters the room.
22:56:08tarcieriheadius: heh, never mind, I was just confused :/
22:56:17headiusok :)
22:56:52ljulliardrbrain: don't you think there is a bug in kernel/core/global.rb, method def []=(key, data)
22:56:55ljulliar?
22:57:14drbrainwhat bug?
22:57:25jewel enters the room.
22:57:47ljulliarwhen there is a setter associated, the setter is called but the result of th block is not stored in @internal[key] as it should, I think.
22:58:06ljulliarline 62
22:58:13drbrainyes, that is a bug
22:58:23vjmp leaves the room.
22:58:42ljulliarOk I'll make the change
23:02:57eventualbuddha enters the room.
23:05:00headiusdoes the child vm get access to the parent ID somewhere?
23:05:10smparkes1 leaves the room.
23:07:32vjmp enters the room.
23:07:39headiusI don't see that in the code
23:09:41cremes leaves the room.
23:11:04headiuspastie
23:11:18pastiehttp://pastie.org/236049 by headius.
23:12:09headiuswow, that pasted weird
23:12:09headiushttp://pastie.org/236050
23:17:26benny leaves the room.
23:19:41headiustarcieri: are you using the VM stuff?
23:19:42fbuilesv leaves the room.
23:21:55benny enters the room.
23:22:04michalw enters the room.
23:23:10tarcieriheadius: Not for anything practical... yet
23:23:15headiusmm ok
23:23:21headiusare there any more docs than the code?
23:23:23tarcieriVMActor is the only thing using it right now, afaic
23:23:27headiusthere's no rdoc or anything in there
23:23:30tarcierierr, afaik
23:23:39tarcieriin what? Rubinius::VM?
23:24:00heycarsten leaves the room.
23:24:10headiusyeah
23:24:24tarcierinope
23:24:33tarcieriI figured it out either by looking at the code or asking evan :/
23:25:29headiusheheh
23:25:40headiuswell so far the jruby version is all ruby
23:25:53headiusso I'll probably try to doc it a bit more
23:26:21tarcierioh nice, doing MVM for JRuby?
23:26:35tarcieriwhatever became of the whole standard MVM interface initiative thingy?
23:26:37headiusyeah, figured I'd finally write it up
23:26:41headiusit's still going
23:26:52headiusI think ko1 et al have had their hands full on the native side
23:27:01headiusnot a lot of work done by them on the API
23:27:50tarcieriI see
23:29:00nathansobothis is probably an oft-asked question, so excuse me if i'm annoying, but how close is rubinius to being faster than 1.9
23:29:26drbrainit'll be a while
23:29:42nathansobolike a year?
23:29:55nathansobowe're thinking of porting to erlang but i really don't want to
23:30:16drbrainI would estimate no sooner than six months
23:32:49headiusporting what?
23:33:23headiusand you don't want to use 1.9 because of weak concurrency capabilities?
23:33:57tarcieriConcurrency on 1.9 is just fine, depending on your needs
23:34:21headiuswell, except that it isn't concurrent unless you twiddle stuff at the C level
23:35:09massactual concurrent concurrency is overrated
23:35:13tarcieriIf you can fit what you're doing into a coroutine model then Revactor provides nice concurrency on 1.9
23:35:27nathansoboour application
23:35:46nathansoboit's super concurrent
23:35:55headiusmass: well...peopl